Available online at www.sciencedirect.com
Procedia Engineering 45 (2012) 435 – 441
2012 International Symposium on Safety Science and Technology
Study on ignition probability of flammable materials after leakage
accidents
ZHU Changlonga,*, JIANG Junchenga, YUAN Xiongjunb
a
School of Urban Construction and Safety Engineering, Nanjing University of Technology, Nanjing 210009, Jiangsu, China;
b
School of Environment and Safety Engineering, Changzhou University, Changzhou 213164, Jiangsu, China
Abstract
It is a key step of quantitative risk analysis (QRA) to estimate ignition probability of flammable materials after leakage accidents. This
paper reviews the available literature and expert opinion on how to evaluate and determine ignition probability value, and it was
detailedly discussed on the main influencing factors of ignition probability, including flammable material properties, mass flow rate of
flammable materials spillage, ignition resources and ignition controls. Moreover, the operational and practical ignition probability value
could be estimated from the all way of classifications of flammable material, mass flow rate, ignition resources, hazardous areas and
ignition prevention and control measure. Furthermore, the more practical ignition probability model was put forward that the ignition
probability was the maximum value of the probability decided by material properties (PMP), mass flow rate(PQ) and ignition resources(PIS)
with the factor of preventing and controlling ignition (KIC). Finally, the further research was proposed to assign some feasible weigh
factors of the ignition probability for flammable materials after leakage accidents.
© 2012 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of the Beijing Institute of
Technology. Open access under CC BY-NC-ND license.
Keywords: ignition probability; flammable material; leakage accident; QRA
1. Introduction
Ignition probability refers to the ignited probability of some flammable materials after leakage. Estimation of ignition
probability is a key step in the quantitative risk analysis (QRA) for installations where flammable liquids and gases are
stored or transported [1–2]. Ignition can be separated into the immediate ignition and delayed ignition. Immediate ignition
can be considered as the situation where the flammable materials are ignited immediately after leakage accidents by autoignition or accident ignition source. Delayed ignition is the result that the release and diffusing flammable gas cloud is
ignited by other ignition sources apart from the release point. It is assumed to result in flash fires or explosions, and also to
burn back to the source of the leak resulting in a jet fire or a pool fire. None of the existing accident databases makes a clear
distinction between delayed and immediate ignition [3]. Sometime they can be distinguished in principle, but it is not
always in practice. So, in this paper, ignition probability data includes both immediate and delayed ignition probability
without some declaration. At present, QRA for accident consequences puts much emphasis on the influencing radius of
accident, and some research on the determination of ignition probability value is ignored [4–5]. Therefore, it is necessary to
comprehensively study on the influencing factors of determining the value of ignition probability and the ignition
probability models according to some different scenarios.
* Corresponding author. Tel.: +86-13645156946; fax: +86-025-83236946.
E-mail address: [email protected]
1877-7058 © 2012 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2012.08.182
436
ZHU Changlong et al. / Procedia Engineering 45 (2012) 435 – 441
2. Influence factors of determination of ignition probability value
In the study by A Ronza et al (2007), a comprehensive collection of ignition probability data found in specialized risk
assessment literature was listed. Nineteen studies were found that contained quantitative ignition probability data of
flammable spills. According to these studies, the factors that influence ignition probability are: (1) flow rate or amount
released: the greater the release, the larger the area covered by the ignitable cloud and the higher the probability of it finding
an ignition source; (2) the substance released: the more volatile and flammable the material, the more likely the ignition; (3)
the characteristics of the surroundings and the general conditions of the leak, on which the number and effectiveness of
potential ignition sources depend. In some literatures [1–2, 6–7], delayed ignition probability is considered to be
significantly affected by material released, cloud size, release duration and cloud location and number of available ignition
sources. The following will focus on analysis and comparation with much scientific research on ignition probability models
and expert opinion. Some view about these factors of determination of ignition probability will be put forward.
2.1. Flammable material properties (MP)
Some benchmarks on flammable material properties include flash point temperature (FPT), ignition temperature (IT),,
auto-ignition temperature (AIT), minimum ignition energy (MIE), the lower flammable limit (LFL) and the upper
flammable limit (UFL). Flammable material properties can be regarded to the function of these variables. Ignition will occur
only if a gas or vapour cloud finds an ignition source within its flammable range (The concentrations between LFL and UFL
constitute the flammable range). FPT is a key parameter in the hazard classification of flammable materials and in
determining their likelihood of ignition. For example, a hydrocarbon fluid which has a flash point below ambient
temperature when released to atmospheric pressure has the potential to be ignited. A kind of gas or vapour with low flash
point can be readily ignited by a small flame or spark, whereas a kind of non-volatile liquid with high flash point would
require a much more intensive heat source.
How the MP influences on the determination of ignition probability is discussed in Purple book, BEVI manual and some
other literatures as following: According to Purple book [1], MP is a key factor to determine ignition probability. The
probability of direct ignition for stationary installations is given in Table 1.
Table 1. Probability of direct ignition for stationary installations
Source
Substance
Continuous/(kg·s-1)
Instantaneous/kg
K1-liquid
Gas, low reactive
Gas, average/high reactive
<10
<1 000
0.065
0.02
0.2
10–100
1 000–10 000
0.065
0.04
0.5
>100
>10 000
0.065
0.09
0.7
In the study by Mike Moosemiller (2011), the probability of immediate ignition (Pimm.ign) depends upon both the potential
for auto-ignition and the potential for static discharge. In this model, the former is related to the release temperature (T)
relative to the auto-ignition temperature (AIT), and the latter to the minimum ignition energy (MIE) and “release energy”
for the material being released. The “release energy” may be considered a function of the process pressure, or release rate,
or some other surrogate parameter. These variables are taken into account, resulting in the following formula
Pimm.ign
Pformer
Platter
[1 5000e
9.5 ( T / AIT )
] [0.0024 (145.04 P )1/ 3 / (M IE ) 2 / 3 ]
(1)
where AIT and T are in degrees Fahrenheit, P is in MPa, and MIE is in mJ. The following constraints place on this equation.
A minimum value of 0 is allowed for T. For T/AIT<0.9, Pformer=0; For T/AIT>1.2, Pformer=1. Pimm.ign cannot be greater than 1.
A Ronza (2007) analyzed tens of thousands of records of hydrocarbon spills by HMIRS (a database about hazardous
materials incident reporting system). The relationship between the statistical data of ignition probability and the average
flash temperature of the hydrocarbon spilled is established. Fig.1 describes how ignition probabilities vary as the average
flash point of the hydrocarbon blend. Furthermore, it is seen that the average ignition probability decreases with the flash
temperature of the hydrocarbon spilled.
ZHU Changlong et al. / Procedia Engineering 45 (2012) 435 – 441
437
0.12
LPG
Ignition Probability
0.1
0.08
0.06
0.04
light fractions
0.02
0
kerosene/jet fuel
crude oil diesel oil/gas oil
fuel oil
-100
-50
0
50
100
Flash temperature(℃ )
Fig. 1. Relationship between the average HMIRS ignition probability data and the average flash temperature of the hydrocarbon spilled.
BEVI manual [2] provides a more scientific method of determining ignition probabilities, which gives the different
values according to the categories of some flammable materials. The definition of the substance category is given in Table 2,
which derived from WMS (environmentally hazardous substances act).
Table 2. Classification of flammable substances
Substance
category
WMS
category
Limits
Category 0
Extremely
flammable
Liquid substances and preparations with a flash point lower than 0 and a boiling point (or the start of a boiling range)
less than or equal to 35 °C; Gaseous substances and preparations that may ignite at normal temperature and
pressure when exposed to air
Category 1
Highly
flammable
Liquid substances and preparations with a flash point below 21 °C, which are not, however, extremely flammable
Category 2
Flammable
Liquid substances and preparations with a flash point greater than or equal to 21 °C and less than or equal to 55 °C
Category 3
Liquid substances and preparations with a flash point greater than or equal to 55 °C and less than or equal to 100 °C
Category 4
Liquid substances and preparations with a flash point greater than 100 °C
The probability of direct ignition depends on the type of installation (stationary installation or transport unit), the
substance category and the outflow quantity. For example, the values for stationary installations are given in Table 3.
Table 3. Probability of direct ignition for stationary installations
Source term-continuous/(kg·s-1)
Source term-instantaneous/kg
Probability of direct ignition
<10
<1 000
0.2
10–100
1 000–10 000
0.5
>100
>10 000
0.7
<10
<1 000
0.02
10–100
1 000–10 000
0.04
>100
>10 000
0.09
Category 1
all flow rates
all quantities
0.065
Category 2
all flow rates
all quantities
0.01
Category 3, 4
all flow rates
all quantities
0
Substance category
Category 0
Average/high reactivity
Category 0
Low reactivity
438
ZHU Changlong et al. / Procedia Engineering 45 (2012) 435 – 441
The essential factor of MP influencing the ignition probability is mainly inclusive FPT, AIT and MIE. Considering of
some inherent positive correlations among the flammable properties, it is appropriate to choice the variable of FPT or MIE
as the independent variable of the ignition probability function. For example, when we use the suggested formula (1) to
calculate the ignition probability of some flammable materials, the formula can be simplified to the correlation of just about
MIE because T/AIT is less than 0.9 in most cases. In order to balance the defect between expert opinion on ignition
probability and statistical analysis of the existing accident database, it will have better objectivity and operability to
determine the ignition probability of material according to the categories of some flammable materials, just like the BEVI
Manual using the classification of substance to determine the value of ignition probability. At the same time, this method is
a more preferable especially when the accident scenes are not detailed.
2.2. Mass flow rate of flammable materials leak (Q)
Amount spilled of flammable materials can be considered to the function of flow rate, the duration time of leak, the
diameter of leakage location and the pressure difference between inner and outer. It tends to determine the value of ignition
probability using the mass flow rate rather than other variables. Above a few tables several methods of determining the
value of ignition probability are directly given through the amount spilled of flammable materials. In addition, Cox et al
(1990) suggested a correlation for the probability of ignition based on mass flow rate, i.e. for continuous rather than
instantaneous releases. It is assumed that the probability of ignition is proportional to a power of the mass flow rate, and the
constant of proportionality and the power are then set from a few data points. If the mass flow rate is denoted by Q (kg/s),
the probability of ignition for a particular scenario is approximately given by
p aQ b
(2)
where values of the coefficients a and b are estimated for a few scenarios. A Ronza (2007) used the similar methodology to
analyze the correlations of ignition probability and the mass flow rate such as LPG, light fractions, crude oil kerosene, jet
fuel diesel oil and gas oil. The coefficients a and b resulting from the fit curves are obtained respectively. A guidance for
quantitative risk assessment in the petrochemical plant (2007, China Petrochemical Press) also advised the empirical
formula of determining the value of delayed ignition probability as following [8]
Pgas.delay.ignition
(e
4.16
m0.642 ) (e
2.995
Pliquid.delay.ignition
(e
4.333
m 0.38 )
(3)
m0.392 ) (e
2.995
m0.38 )
(4)
where m is the mass flow rate, kg/s.
A report (2010, The International Association of Oil and Gas Producers) illustrated a better method of determining the
value of ignition probability, which calculated the value of ignition probability with different mass flow rates by means of
using the end point value of four sections of broken lines[9–10]. The data used to generate the lines are shown in Table 4.
This specifies the release rates (RR, kg/s) and ignition probabilities (P) relating to each of the points bounding the segments
as indicated in Fig.2.
Table 4. Data for Look-up Correlations
Scenario type
Point 1
Point 2
Point 3
P
Small plant gas LPG
0.1
0.001
1.00
0.002 5
Small plant liquid
0.1
0.001
1.00
0.002 4
Large plant gas LPG
0.1
0.001
1.00
0.002 5
260.00
0.65
Large plant liquid
0.1
0.001
1.00
0.002 5
109.99
0.13
Large plant congested gas LPG
0.1
0.001
1.00
0.002 5
70.00
0.43
Tank gas LPG plant
0.1
0.001
1.00
0.001 2
102.84
1.00
Tank gas LPG storage only industrial
0.1
0.001
1.00
0.001 2
100.00
RR/(kg·s-1)
P
Point 4
RR/(kg·s-1)
P
RR/(kg·s-1)
P
3.00
0.01
498.99
0.60
100.00
0.10
325.03
0.70
0.23
988.11
1.00
10.00
0.02
52551
0.50
25.55
0.002 4
RR/(kg·s-1)
Tank gas LPG storage only rural
0.1
0.001
1.00
0.001 2
Offshore process liquid
0.1
0.001
100.00
0.02
Offshore process gas typical
0.1
0.001
3.00
0.01
37.01
0.04
Tank liquid-diesel and fuel oil
0.1
0.001
1.00
0.001
7.00
0.001 4
439
ZHU Changlong et al. / Procedia Engineering 45 (2012) 435 – 441
The logarithm of the required probability is obtained by the following formula
lg Pign
lg Pign.lower
(lg Q lg Qlower )(lg Pign.upper
lg Qupper
lg Pign.lower )
lg Qlower
where Pign is the required ignition probability corresponding to release rate
(5)
Q , Pign.lower is the ignition probability at a
release rate of Qlower (the lower bound of the relevant curve section) , and Pign.upper is the ignition probability at a release rate
Qupper (the upper bound of the relevant curve section).
ignition probability
(P)
of
Point 4
Point 3
Point 1
Point 2
Release Rate
Fig. 2. Relationship of release rate (RR) and ignition probability (P)
It tends to determine the value of ignition probability using the flow rate rather than other variables. It is more scientific
to determine the value by dividing into interval segments in according with the mass flow rate on some typical scenarios
because the relatively accurate statistical value of these end points can guarantee the relatively correct result of different
flow rate. The operability and objectivity will be explicit, and it is convenient to construct a spreadsheet or some program to
carry this out.
2.3. Ignition sources (IS)
Ignition source is a key factor determining the value of ignition probability. Rew & Spencer (1997) gave a framework
ranking system, derived from a number of some references, which was given in Table 5.
Table 5. Framework ranking system for ignition sources
Category (strength of source)
Examples of ignition sources
Ignition potential
Certain
Pilot light; open flare
p=1
Strong
Electric motors; hot work
p>0.5
Medium
Vehicles; faulty wiring
0.5>p>0.05
Weak
Electrical appliances; mechanical sparks
p<0.05
Negligible
Intrinsically safe equipment; radio frequency sources
p=0
In the study of Purple book, the probability of delayed ignition caused by an ignition source can be modeled as
P (t )
Ppresent (1 e
t
)
(6)
where P(t) is the probability of an ignition in the time interval 0 to t(-); Ppresent is the probability that the source is present
when the cloud passes(-); is the ignition effectiveness, s-1; t is time, s. The ignition effectiveness, , can be calculated
given the probability of ignition for a certain time interval. Fig.3 shows the probability of ignition for a time interval of one
minute for a number of sources.
In the study of Mike Moosemiller (2011), the similar theory was used to propose a calculation formula of the ignition
sources influencing factor (Mduration), which is given below [11].
M duration
1 (1 s 2 ) e
(0.015 s ) t
(7)
where strength “s” is denoted to be probability of ignition in one minute. The equation (7) takes a modified form of the
equation (6).
440
ZHU Changlong et al. / Procedia Engineering 45 (2012) 435 – 441
Not only the ignition sources are taken into account, but also the size and duration time of the gas cloud that continuously
covers the ignition source influence on the probability of ignition. Simmons (1974) conducted a survey of 59 incidents of
ignition of clouds of LNG or LPG resulting from accidental spills due to transportation [12]. For these, the size of the cloud
when ignition occurred was estimated and fitted the probability of ignition as a function of cloud area. By considering in
depth, for a flammable cloud of area A, containing a random distribution of ignition sources with parameters , p and a, the
probability Q(t) of no ignition at time t in a given type of scenario is [6–7]:
Q (t )
exp(
A{1 (1 ap )e
pt
})
(8)
where μ is the average number of ignition sources per unit area; p is the ignition potential of a source (0–1); a is the rate of
activation of the source; is the proportion of time that the source is active.
Point Source
Line Source
Area Source
electric train
diesel train
transmission line:
0.2 per 100 m
pleasure craft
fishing vessel
road /railway:
as for the average
traffic density
and the ignition
effectiveness of
a single vehicle
ship
indoor boiler
outdoor boiler
indoor furnace
chemical plant/oil
refinery: 0.9 per
site
heavy industry: 0.7
per site
light industrial
warehousing: as for
population
outdoor furnace
flare
motor vehicle
Probability of ignition in one minute(strength"s")
Probability of ignition in one minute
Fig. 3. Probability of ignition for a time interval of one minute for a number of sources.
The existing ignition resources and the contact with the gas cloud are the essential conditions resulting in ignition. The
theory model (equation (8)) built is correct, but the calculation in practice is very difficulty because of the uncertain
distribution and appearance of the ignition resources. So it is better ways to classify the ignition sources and use the
recommended equations by some experts such as the equation (7). Due to the influence of ignition sources, semiquantitative estimation of ignition probability is the preferable strategy to qualitative estimation and theoretical calculation.
2.4. Ignition controls (IC)
The key approaches to controlling ignition sources are: (1) Hazardous area classification, which essentially, defines the
areas where flammable substances may occur and determines the controls over the ignition sources in design and normal
processes in these areas. (2) Permit-to-work systems, which ensure that any non-routine work is carried out in a safe manner
and determine the controls over the temporary and transient ignition sources. In most cases, it is often more appropriate to
define the parameters for a “typical” level of ignition control and then to use the factors to determine other control levels
accordingly. An estimation of ignition probability is to rank system of ignition source controls just as Table 6 [6].
Certainly, there are other measures to decrease the ignition probability such as the effectively ventilation, the essential
safety design of some electrical equipments, the strict management system, etc.
3. Ignition probability modeling
Some main factors determining the value of ignition probability are discussed above. Equation (8) is the more suitable
model in quantitative calculation. But the calculation is very difficulty in practice because of the uncertain distribution and
appearance of the ignition resources. During the course of carrying out quantitative risk analyses, the model is lack of
practicability. So the new model must be yielded to the practical availability. In view of the nonlinear relationship among
the factors of MP, Q, IS and IC and these variables should be considered comprehensively in the new model, the more
practical method is advised as following: PIP=maximum(PMP, PQ, PIS)×KIC, where PIP is the ignition probability, PMP is the
probability decided by material properties, PQ is the probability decided by mass flow rate, PIS is the probability decided by
ignition resources, and KIC is the factor decided by the ignition controls.
ZHU Changlong et al. / Procedia Engineering 45 (2012) 435 – 441
441
Table 6. Framework ranking system of ignition source controls
Type
Probability of ignition, in presence of
flammable substance
Ignition control
Quantitative
Qualitative
Ideal
0
None
Excellent
0.1
Minimal
Well designed and maintained – ignition only arising from rare events (e.g.
unforeseen failure of equipment)
Typical
(good)
0.25
Limited
Designed to meet standards and maintained regularly—ignition eliminated in
normal operation, but potential for failure of systems or changing
circumstances to result in occasional ignition source
Poor
0.5
Poor
Does not meet precise standards and poorly maintained—significant potential
for ignition sources to occur
None
>0.5
None
No adherence to standards and little maintenance—significant potential for
ignition sources to occur
Design and maintenance ensures no ignition source at any time
By the analysis of the influence factors or models of ignition probability, some suggestions are given as following: (1)
For the PMP, it is feasible and necessary to categorize the flammable materials referred by WMS and BEVI. (2) For PQ, it is
a good strategy to determine the value by dividing into interval segments in according with the mass flow rate on some
typical scenarios, which is referred by mathematical functions drawn from some look-up correlations. (3) For PIS, it can be
calculated directly by equation (8) if the gas cloud area, duration time and the detail distribution of the existing ignition
resources are known. But in most cases, the accidents scenarios are unknown. Therefore, it is better ways to classify the
ignition sources and use the recommended equations by some experts such as equation (7). Semi-quantitative estimates of
ignition sources are the preferable strategy to qualitative estimation and theoretical calculation. (4) For KIC, it is also
determined by classification of some control measures including management, electrical equipments, etc. KIC is the
modification parameter. (5) In order to make the more conservative result, the function of ignition probability is the
maximum value of (PMP, PQ, PIS)×KIC rather than the minimum.
4. Conclusions
(1) The main influence factors of ignition probability are inclusive of the flammable material properties, mass flow rate
of flammable materials spillage, ignition resources and ignition controls. (2) Estimation of ignition probability is a key step
in the quantitative risk analysis (QRA). Classifications of flammable material, the mass flow rate, ignition resources,
hazardous areas and ignition controls measures are put forward, which makes the determination of ignition probability
operational and practical. (3) The more practical ignition probability model is advised as following: PIP=maximum (PMP, PQ,
PIS)×KIC. (4) After classification of some factors, the assigned feasible weigh according to contribution to the ignition
probability will be further researched, which brings convenience to construct the a few variables function model.
References
[1] Committee for the Prevention of Disasters, 2005. Guide lines for quantitative risk assessment (Purple book), CPR 18E, PGS, Hague.
[2] BEVI, 2009. Reference manual BEVI risk assessments version 3.2, National Institute of Public Health and the Environment (RIVM), Bilthoven.
[3] Ronza, A., Vílchez, J.A, Casal, J., 2007. Using transportation accident databases to investigate ignition and explosion probabilities of flammable spills,
Journal of Hazardous Materials 146, p. 106.
[4] Yuan, X.J., Zhu, C.L., 2011. Study on taking values of ignition probabilities in hazardous chemical leakage accidents, China Safety Science Journal 21,
p. 39.
[5] Bai, Y.Z., Yu, A.F., 2010. Determination of ignition source and ignition probability of leaking accidents in petrochemical processes, China Safety
Science Journal 20, p. 61.
[6] Daycock, J., Rew, P.J., 2004. Development of a method for the determination of on-site ignition probabilities, HSE Research Report.
[7] Rew, P.J., Spencer, H., Daycock, J., 2000. Off-site ignition probability of flammable gases, Journal of Hazardous Materials 71, p. 409.
[8] China Petroleum & Chemical Corporation, Qingdao Institute of Safety Engineering, 2007. A guidance for quantitative risk assessment in the
petrochemical plant, China Petrochemical Press, Beijing, p. 54.
[9] Energy Institute, 2006. Ignition probability review, model development and look–up correlations, Energy Institute, London.
[10] International Association of Oil &Gas Producers, 2010. Risk assessment data directory-Ignition probabilities, No.434–6.1.
[11] Moosemiller, M., 2011. Development of algorithms for predicting ignition probabilities and explosion frequencies, Journal of Loss Prevention in the
Process Industries 24, p. 259.
[12] Simmons, J.A, 1974. A risk assessment of storage and transport of liquefied natural gas and LP-Gas.